hamachi@ucbkim (03/02/85)
From: hamachi@ucbkim (Gordon Hamachi)
Mike at Megamax called me concerning my recent compiler evaluation. He
pointed out one outright error, and one misleading piece of information
in one of my previous messages concerning the Megamax compiler. If you
had decided not to buy their compiler because of some negative things I
said, please read this message and reconsider.
First, I was just plain wrong when I said that there was an error assigning
a long to an int. I forgot to use "l" as in "%ld" in my printf statement,
so the value my test program printed was in error. I fixed this and
everything now works fine.
Second, I found the stack overflow problem in the OLD version of the
Megamax compiler. The folks at Megamax were aware of the problem and have
fixed it in their new (just released?) version. If you have the old
version, simply avoid arithmetic expressions within WHILE loops, of the form:
while(tickcount() <= delay*60+startTime)
{
}
As far as I know, all of the other things I said about the Megamax compiler
are true. It compiles the fastest, produces the smallest binaries, and loads
and executes very quickly. The Megamax employees are very friendly and
helpful.
I would not hesitate to recommend this compiler to anyone, particularly if
you need floating point right now, and you would rather not use a unix-like
tty interface. In fact, a friend of mine purchased the Megamax compiler
several weeks ago after asking for my advice.
On the other hand, I have not had very much experience with the other
compilers. I hope to get evaluation copies from Megamax, Manx, Consulair,
Hippo, and Softworks so I can do a more careful (!) comparative evaluation.
--Gordon Hamachi